Training Your AI Agent
After activating your AI agent, the most important step is testing and training it through real call scenarios.
Make several test calls to simulate different maintenance requests and caller behaviors.
This helps refine how the AI agent responds.
Reviewing Call Activity
To review interactions:
- Navigate to Activities.
- You can also access ratings by agent. Within any agent details, you can view their Recent Activities.
- Open a call conversation
- Review the transcript, summary, and outcome
Rating Calls
Each conversation can be rated to help improve the AI agent’s performance.
Three rating options are available:
Rating | Meaning |
😊 Happy Face | The AI agent handled the call well. |
😐 Neutral Face | The call could have been handled better, but no specific correction is provided. |
☹️ Sad Face | The AI agent did not handle the call correctly and needs additional guidance. |
When selecting the sad face, you will be prompted to provide feedback describing:
- What the AI agent did incorrectly
- What the agent should have done instead
This feedback helps improve the agent’s future responses.
Improving Responses with Knowledge
If the AI agent provides incorrect information, you can improve its responses by updating the Knowledge section.
Add the correct information the AI agent should reference when answering similar questions.
Important:
The knowledge section provides information for the agent to reference, but it does not control agent actions or instructions.
✅ Tip:
Regularly reviewing conversations and providing feedback will help your AI agent improve over time and better handle real maintenance requests.